基於魚眼鏡頭具有大約180 °甚至更大視場角之特性,相較於一般相機鏡頭可獲得更廣泛的空間資訊。透過視場角廣泛之優勢,在實務作業上使用魚眼鏡頭可大幅度減少蒐集空間資訊所需拍攝影像張數,進而提升任務效率。另外在某些拍攝距離受限之區域,如室內空間與隧道等,魚眼鏡頭尤能顯現其優勢。 然而魚眼鏡頭伴隨而來之缺點為影像嚴重變形,故若未經過審慎的率定過程以及良好的網形配置,魚眼影像則無法應用於後續高精度任務。 據此,為使用魚眼鏡頭達成毫米定位等級的高精度三維定位任務,勢必需要次像元的率定成果以及探討合適的相機幾何、控制條件與解算過程。然而就目前參閱的文獻,並未有文獻提供一套可參考的規則或是條件能夠有效率並成功地達到任務目標。故本研究設計以長廊狀場域為例,在高精度的內方位參數與嚴密的物像對應關係下,透過布設數量充足且分布良好之定位毫米等級的控制點與檢核點並進行光束法平差,探討物空間mm等級三維定位所需具備的控制品質(包括精度及分布)、像點品質與攝像條件。
With a field of view (FOV) of approximately 180 ° and even larger, fisheye lens can obtain a wider range of spatial information than general camera lens. Through fisheye lens’ wide FOV, the image acquisition can be made by invloving less number of images and thus conducted in a much efficient way to effectively support geo-spatial information collection for those areas restricted in the short-range indoor or tunnel-like environment. Yet the significant distortion of the image through fisheye lens prevents such imaging system from quality applications should the precise calibration and nice network can not be made possible. Therefore, in order to cope with the highly accurate 3D positioning task at the mm level of the close-range scene environment by using fisheye lens, one needs to obtain at least the sub-pixel level calibration result and arranges appropriate camera geometry, control condition and computation processing. However, to the authors’ knowledge, there is no referable rules or conditions giving a fluent and successful operation guidance. For that, we design a corridor-like field with sufficient number of well-distributed mm-level control points and check points and perform bundle adjustment to analyze the influential factors and comb the knowledge or experience of how to configure control quality (including quantity and distribution), number of images, camera condition and intersection geometry for achieving mm-level 3D positioning accuracy when using fisheye lens in mapping such similar environment.